Photo vector extraction using Deep Learning techniques for Cognitive Map Index calculation
-
M. Jaén-Vargas [1] ; M. Fernández Cerezo [1] ; K. Reyes Leiva [1] ; F. Ortiz Gomez [1] ; A. Cobo [1] ; J. Serrano Olmedo [1]
-
[1] Universidad Politécnica de Madrid
Universidad Politécnica de Madrid
Madrid, España
-
- Localización: XXXVIII Congreso Anual de la Sociedad Española de Ingeniería Biomédica. CASEIB 2020: Libro de actas / Roberto Hornero Sánchez (ed. lit.), Jesús Poza Crespo (ed. lit.), Carlos Gómez Peña (ed. lit.), María García Gadañón (ed. lit.), 2020, ISBN 978-84-09-25491-0, págs. 89-92
- Idioma: inglés
- Enlaces
-
Resumen
The main purpose of this project is to automate the manual mechanism of the acquisitioning of 3 vectors from the images taken of an experiment executed by impaired people. These vector values are implied to calculate de CMQ index which intervenes to obtain how well a person mentally understands their exterior space by creating cognitive maps. The index mentioned was obtained in a more efficient and less time-consuming process. The system includes a smartphone application that aims users to identify Point of Interests (POI) in a scenario previously virtualized. Importantly, Deep Learning techniques were applied to automate the process, principally, two supervised learning approaches: regression and classification. While classification helps to improve the model, the regression approach is closer to the expected result. However, neither of the two types of algorithms provides the desired results since the model presents a high error in the prediction, although in turn, they show a clear capacity for improvement. Regarding the technologies used in the study, Python is the main language for coding besides Google Collab where compiling models in the cloud. The results indicate that the tested model represents a first approximation that requires a greater number of samples and further development to obtain a more accurate prediction and be implemented, but the path is appropriate.
